Glucose and Tolbutamide Induce Apoptosis in Pancreatic β-Cells

High concentrations of glucose are considered to be toxic for the pancreatic β-cell. However, the mechanisms underlying β-cell dysfunction and resulting cell death are not fully characterized. In the present study we have demonstrated that incubation of pancreatic islets and β-cells from ob/ob mice and Wistar rats with glucose induced a process of apoptotic β-cell death, as shown by DNA laddering, TdT-mediated dUTP-biotin nick end-labeling (TUNEL) technique, and by using DNA-staining dye HOECHST 33342. The obtained results show that the percentage of apoptotic cells was dependent on glucose concentration, being minimal at 11 mm glucose. At a concentration of 100 μm, aurintricarboxylic acid, an inhibitor of endonuclease activity, almost completely inhibited apoptosis triggered by 17 mm glucose. We have also shown that long term incubation with 100 μm sulfonylurea, tolbutamide, triggered apoptosis in pancreatic β-cells. The process of β-cell death induced by high glucose concentration and tolbutamide were Ca2+-dependent, because introduction to the culture medium of 50 μm D-600 or 200 μmdiazoxide, which blocked glucose- and tolbutamide-induced [Ca2+] i increase, inhibited apoptosis. Thus, this study shows for the first time that high glucose concentrations and tolbutamide induce apoptosis in pancreatic β-cells, and that this process is Ca2+-dependent.


EXPERIMENTAL PROCEDURES
Materials-RPMI 1640 medium and fetal calf serum were obtained from Life Technologies, Inc. IL-1␤, tolbutamide, bovine serum albumin, aurintricarboxylic acid (ATA), diazoxide, palmitate, and oleate were purchased from Sigma. The fluorescent dyes fluorescein isothiocyanate (FITC), propidium iodide (PI), HOECHST 33342 (HO 342), and fura-2 acetoxymetyl ester were from Molecular Probes. The kit for TUNEL reaction was received from Boehringer Mannheim. All other reagents were of analytical grade and obtained from MERCK.
Isolation of Pancreatic Islets and ␤-Cells-10 -12-month-old obese (ob/ob) mice, obtained from a local colony, as well as 2-3-month-old Wistar rats were used. Pancreatic islets from ob/ob mice contain approximately 90% ␤-cells (13). Islets were isolated by collagenase digestion and dispersed into small ␤-cell clusters in Ca 2ϩ -and Mg 2ϩ -deficient medium, as described previously (14). Cells were cultured on glass coverslips in plastic Petri dishes for 40 h in RPMI 1640 medium containing different concentrations of glucose and other test substances, 10% (v/v) fetal calf serum, 100 IU/ml penicillin and 100 g/ml streptomycin. Fatty acids (sodium salts) were dissolved in 95% ethanol before being added to the culture media, as described previously (15). The final concentration of ethanol in the medium was 1.25% (v/v) at the concentration of 0.25 mM palmitate and 0.125 mM oleate. Control incubations with 1.25% ethanol did not affect the apoptotic process.
Fluorescence Microscopy Analysis of Cell DNA Staining with HO-ECHST 33342-The percentages of dead cells were assessed in single cell preparations on glass coverslips after a culture period of 40 h. For this purpose, pancreatic ␤-cells were exposed to the DNA binding dye HO 342 at a concentration of 20 g/ml for 10 min at room temperature. HO 342 freely passes the plasma membrane of intact cells as well as damaged cells and stains DNA blue. Morphological changes corresponding to dead cells were visualized using an inverted fluorescence microscope with a mercury lamp in combination with excitation filter 340 -380 nm and emission filter Ͼ430 nm.
DNA Fragmentation Analysis by Gel Electrophoresis-Pancreatic islets were cultured 40 h in a medium containing different concentrations of glucose and other substances. After this period islets were dispersed into a suspension of ␤-cells as described above. Aliquots from the cell suspensions containing 1 ϫ 10 6 cells were disrupted by resuspension in ultrapure water containing 10 mg/ml RNase and incubated for 20 min at room temperature. Loading buffer was added to this suspension.
Electrophoresis was run with a constant voltage of 20 V overnight and then 90 V for 1 h on 1.8% agarose gels containing proteinase K. The mixture of pBR 328 DNA-BglI and pBR 328 DNA-HinfI was used as a marker. The separated DNA was stained with ethidium bromide, visualized by UV light, and photographed.
TUNEL Labeling of Pancreatic ␤-Cells and Double Staining for Confocal Microscopy-TUNEL technique was used to detect DNA strand breaks in situ. After 40 h of incubation, pancreatic ␤-cells were washed with phosphate-buffered saline and fixed in 80% methanol at 4°C. After rinsing with phosphate-buffered saline, cells were covered with 100 l of reaction mixture for TUNEL enzymatic reaction and were incubated at 37°C for 1 h. The reaction was stopped by adding a buffer containing 300 mM NaCl and 30 mM sodium citrate. Apoptotic cells were determined using the combination of two fluorescent dyes, FITC and PI. Cells doubled stained with FITC and PI were fixed on glass slides with 50% glycerol in phosphate-buffered saline. Fluorescence was monitored with a Leica TCS NT laser-scanning confocal microscope (Leica Lasertechnik GmbH, Heidelberg, Germany), with excitation from the 488-nm line of an argon/krypton laser. Fluorescence emission was detected with a bandpass filter (from Chroma Technology Corp.) centered at 530 nm for FITC and above 590 nm for PI. The pinhole was set to give a confocal section thickness of 1 mm, using a 25 ϫ 0.75 PL Fluotar oil-immersion objective.
Presentation of Results-Confocal images were processed using a Corel Photo-Paint and Corel Draw program. Data analysis were performed using the program Sigma Plot for Windows (version 4, Jandel Corp.). Given protocols were tested in pancreatic ␤-cells from 3-5 preparations from different animals. The statistical significance between means was assessed by Student's t test for unpaired values.

RESULTS
To evaluate the effect of glucose on pancreatic ␤-cell survival, the cells were incubated for 40 h in RPMI 1640 medium with 0, 3,5,8,11,14,17,20,23, and 27 mM glucose and then examined for nuclear DNA staining with fluorescent probe HO 342. As shown in Fig. 1, the absence of glucose in the culture medium gave rise to pronounced cell death. ␤-Cell death was minimal at 8 -11 mM glucose and increased at higher concentrations ( Fig.  1). Because RPMI 1640 medium with 11 mM glucose was shown to be optimal for culturing of pancreatic ␤-cells, this condition was chosen as the control.
For evaluation of apoptotic changes after exposure to high glucose concentration, analysis of pancreatic islet DNA fragmentation by gel electrophoresis as well as by TUNEL labeling and double staining of pancreatic ␤-cells for confocal microscopy were performed.
The extent of DNA laddering is shown in Fig. 2. There was no detectable DNA laddering in gel electrophoresis after incubation of ob/ob mice pancreatic islets in RPMI 1640 medium with 11 mM glucose (standard culture conditions) for 40 h (Fig. 2, first lane). The addition of 100 units/ml IL-1␤, a well known inducer of apoptosis in ␤-cells (17,18), led to characteristic apoptotic DNA laddering pattern (Fig. 2, second lane) and was used as a positive control for evaluation of apoptosis in our system. The incubation of pancreatic islets in medium containing 17 mM (Fig. 2, third lane) or 27 mM (Fig. 2, fourth lane) glucose during 40 h led to DNA laddering, suggesting that long term exposure to high glucose concentration triggered apoptosis. Interestingly, in this case 17 mM glucose induced more DNA laddering than 27 mM glucose.
It is well known that Ca 2ϩ plays an important role in the regulation of the apoptotic process in multiple experimental models (19 -22). To investigate whether the increase in [Ca 2ϩ ] i induced by glucose was responsible for the activation of the apoptotic process under our experimental conditions we used diazoxide, an opener of K ATP channels, which hyperpolarizes  (Fig. 2, fifth lane) and 27 mM glucose (Fig. 2, sixth lane), suggesting that glucose-induced apoptosis is a Ca 2ϩ -dependent process.
To study whether a [Ca 2ϩ ] i elevation per se may trigger apoptosis in pancreatic islets, the sulfonylurea tolbutamide was used. Tolbutamide induces Ca 2ϩ influx into ␤-cells by closure of K ATP channels and subsequent opening of voltagedependent Ca 2ϩ channels (23). Incubation of pancreatic islets with 100 M tolbutamide for 40 h led to apoptotic changes, as evaluated by DNA laddering (Fig. 2, seventh lane).
To investigate whether the apoptotic changes observed in pancreatic islets also occurred in dispersed ␤-cells, the TUNEL labeling technique with consecutive FITC/PI double staining and subsequent confocal microscopy analysis were used. The results obtained show that after 40 h of incubation with 11 mM glucose, 7.8 Ϯ 1.4% (n ϭ 8) ␤-cells underwent apoptosis and showed nuclear condensation and chromatin compaction (Figs. 4 and 5). Exposure of dispersed ␤-cells to 100 units/ml IL-1␤, which served as a positive control in our studies, resulted in an increased formation of apoptotic cells (Figs. 4B and 5B). A decrease in glucose concentration in the incubation medium from 11 to 5.5 mM (Fig. 4, A and H, and 5A) and an elevation from 11 to 17 or 27 mM also gave an increase in the number of apoptotic cells (Fig. 4, A, C, and E, and 5A). The addition of 200 M diazoxide to the culture medium containing 17 or 27 mM glucose diminished the number of apoptotic cells almost to the control level (Fig. 4, D and F, and 5A) but did not affect the number of apoptotic cells when the culture medium contained 5.5 mM glucose (Fig. 5A). Similar effects were observed when Ca 2ϩ influx into ␤-cells was inhibited using D-600, a blocker of voltage-gated L-type channels (Fig. 5A). Exposure of pancreatic ␤-cells to 100 M tolbutamide raised the number of apoptotic cells (Fig. 4G and 5B), the effect being completely inhibited by 200 M diazoxide (Fig. 5B).
At a concentration of 100 M ATA, an inhibitor of endonucleases that prevents apoptotic cell death in many systems (24), almost completely inhibited high glucose-induced apoptosis (Fig. 5A). The percentage of dead cells in the presence of ATA was significantly higher at 27 mM than at 17 mM glucose ( The results obtained show that this mixture of free fatty acids did not significantly affect the number of apoptotic ␤-cells (Fig. 5A).
As mentioned above, glucose-induced apoptosis of ␤-cells is a Ca 2ϩ -dependent process. In this context, we were interested in investigating whether there were any differences in the pattern of [Ca 2ϩ ] i changes at the control glucose concentration of 11 mM and at 17 mM of the sugar (Fig. 6, Table I). The results show that elevation in glucose concentration from 3 mM to 11 and 17 mM did not significantly affect the maximal increase in [Ca 2ϩ ] i (⌬R max ). However, the average level of [Ca 2ϩ ] i at 10 min after the addition of glucose (⌬R 10 ) as well as the area under the curve for 10 min were significantly higher at 17 mM than at 11 mM glucose. Besides that, more than half of the ␤-cells stimulated by 11 mM glucose had pronounced oscillations in [Ca 2ϩ ] i . At 17 mM glucose, less than 30% ␤-cells were oscillating, and the oscillatory pattern was less pronounced. At basal glucose concentration (3 mM) there were practically no oscillations at all in [Ca 2ϩ ] i (Fig. 6). Preincubation of ␤-cells for 40 h at 17 mM glucose decreased the percentage of cells responding with an increase in [Ca 2ϩ ] i at 17 mM glucose compared with ␤-cells preincubated at 11 mM glucose, 72.4 Ϯ 8.8 (n ϭ 5) and 91.7 Ϯ 4.4% (n ϭ 11), respectively (p Ͻ 0.05).
All experiments described so for were performed using pancreatic ␤-cells from ob/ob mice. The data presented in Fig. 7 and Fig. 8 show that high glucose concentration and tolbutamide as well as IL-1␤ also induce apoptosis in pancreatic ␤-cells from Wistar rats. DISCUSSION The aim of the present study was to investigate whether insulin secretagoges such as glucose and antidiabetic sulfonylurea compounds can induce ␤-cells apoptosis. Our study shows that glucose induces apoptosis of pancreatic ␤-cells from ob/ob mice as wells as normal Wistar rats in vitro. The fact that the endonuclease inhibitor ATA inhibited ␤-cell toxicity induced by 17 or 27 mM glucose further supports the hypothesis that ␤-cells were dying by an apoptotic process. An interesting notion is that there was no direct correlation between the ␤-cell apoptosis and glucose concentration. An increase in glucose concentration from 3 to 11 mM even promoted survival of ␤-cells. However, further elevation of glucose concentration increased ␤-cell apoptosis. Hence, RPMI 1640 medium with 11 mM glucose is optimal for culturing of pancreatic ␤-cells, these results being in agreement with previous studies (26,27). We have also detected more cells dying by apoptosis after culturing in medium with 17 mM compared with 27 mM glucose. The reason for this is not clear and needs further investigations. One possibility is that at 27 mM glucose there are more cells dying by necrotic cell death, which is not associated with the typical ladder-like DNA fragmentation. Experiments with ATA (Fig. 5C) favor this suggestion.
As pointed out above, 11 mM glucose protected the ␤-cell from apoptosis. We were not able to detect apoptotic changes by gel electrophoresis in pancreatic islets incubated in medium with 11 mM glucose. However, we observed a small percentage of apoptotic cells when dispersed ␤-cells were cultured under the same conditions and apoptotic changes were evaluated by TUNEL technique. There are two explanations for the difference in the results obtained. First, techniques using fluorescent dyes are more sensitive, allowing the detection of apoptotic events before DNA fragments become detectable with DNA laddering technique. Second, islet ␤-cells are likely to be more resistant than dispersed cells to the induction of apoptotic changes.
Hoorens et al. (28) have reported that an increase in glucose concentration from 3 to 10 mM promoted the survival of human pancreatic islets in a serum-free medium, the results explained by activating synthesis of proteins that suppress a constitutive apoptotic program at elevated glucose. These data are in accordance with our findings. However, in their study there was no significant difference in the rate of cell death when glucose concentration was increased from 10 to 20 mM. The disagreement between our finding of increased apoptosis at glucose concentrations higher than 11 mM and that of Hoorens et al. (28) may be explained by differences in culture conditions: 40 h in serum-containing medium in our study versus 8 days in serum-free medium in their study and the source of pancreatic islets.
Ca 2ϩ often plays an important role in the regulation of the apoptotic process (19 -21). Sustained [Ca 2ϩ ] i increases can promote endonuclease activation (29) and apoptotic cell death in several cell systems (19 -21, 30). On the contrary, in other cell systems Ca 2ϩ appears not to be directly involved in the apoptotic process, and in some systems an increase in [Ca 2ϩ ] i can even block apoptosis (31). The information on the role of Ca 2ϩ in apoptosis of pancreatic ␤-cells is rather limited. In a previous study we demonstrated that serum from patients with insulindependent diabetes mellitus can activate Ca 2ϩ influx through voltage-gated Ca 2ϩ channels, resulting in the ␤-cell destruction by apoptosis (22). In the present study we have shown that high glucose concentration and tolbutamide induced apoptosis in a Ca 2ϩ -dependent manner. Moreover, we have shown that blocking of the voltage-gated Ca 2ϩ channels by diazoxide-induced repolarization and by D-600 inhibited apoptosis induced by both high glucose concentration and tolbutamide.
It was suggested earlier that oscillations in [Ca 2ϩ ] i may protect against the toxic effects of elevated [Ca 2ϩ ] i in cells (32,33). We have compared the pattern of [Ca 2ϩ ] i handling in pancreatic ␤-cells at different glucose concentrations with the effect of the sugar on ␤-cell survival. The obtained data show that the absence of oscillations in [Ca 2ϩ ] i at 3 mM glucose and the rather low percentage of oscillations at 17 mM glucose correlate with the increased percentage of dead cells at these glucose concentrations. Noteworthy is that the majority of ␤-cells oscillate at the control glucose concentration of 11 mM. Although these are interesting findings, it is too early to state that oscillations in [Ca 2ϩ ] i indeed protect the ␤-cell against glucose-induced apoptosis.
Although our study suggests a role for Ca 2ϩ in apoptosis induced by high glucose concentration and tolbutamide, the reason for the increased number of apoptotic ␤-cells at low glucose concentration (3 and 5.5 mM) still remains unclear. However, experiments with diazoxide and D-600 show that at these concentrations of glucose the ␤-cell apoptosis is not dependent on Ca 2ϩ influx into the cell. The increased rate of ␤-cell apoptosis at 5.5 mM glucose relative to that of 11 mM glucose is  not likely to be accounted for by lack of metabolic fuels, because the addition of free fatty acids to the medium containing 5.5 mM glucose did not overcome apoptosis. These data correspond to what has previously been found by others concerning ␤-cell survival (26 -28). Why is it that the survival of ␤-cells at 5.5 mM glucose, a concentration that the ␤-cell should normally encounter in vivo, is less than at 11 mM glucose? The reason for this is unclear but probably reflects the difference in signaling pathways activated in vitro and in vivo. Under in vivo conditions a number of receptor agonists work in synchrony with glucose, giving rise to a complex signaling pattern, maybe not only leading to activation of insulin biosynthesis and secretion but also to the formation of one or several survival factors protecting the ␤-cell from apoptosis (28). The majority of the results obtained in this study were obtained using preparations of pancreatic islets and ␤-cells from ob/ob mice. Nevertheless, experiments using pancreatic ␤-cells from normal Wistar rats showed the same increase in ␤-cell apoptosis after elevation of glucose from 11 to 17 mM or addition of tolbutamide, indicating that this is a general phenomenon and not restricted to the ob/ob mouse.
We have shown that both a high glucose concentration and the sulfonylurea tolbutamide trigger apoptosis in pancreatic ␤-cells. Our data favor the view that glucose-and tolbutamide-induced apoptosis of pancreatic ␤-cells are Ca 2ϩ -dependent processes.